{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:TNXHPJGWWBQMFMJKFIHHXKZ2CV","short_pith_number":"pith:TNXHPJGW","schema_version":"1.0","canonical_sha256":"9b6e77a4d6b060c2b12a2a0e7bab3a156eaf29eadbefd99146e0769962270013","source":{"kind":"arxiv","id":"2410.10093","version":1},"attestation_state":"computed","paper":{"title":"How to Leverage Demonstration Data in Alignment for Large Language Model? A Self-Imitation Learning Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Chao Cui, Huaisheng Zhu, Mingxiao Li, Teng Xiao, Vasant G Honavar, Yige Yuan","submitted_at":"2024-10-14T02:21:29Z","abstract_excerpt":"This paper introduces a novel generalized self-imitation learning ($\\textbf{GSIL}$) framework, which effectively and efficiently aligns large language models with offline demonstration data. We develop $\\textbf{GSIL}$ by deriving a surrogate objective of imitation learning with density ratio estimates, facilitating the use of self-generated data and optimizing the imitation learning objective with simple classification losses. $\\textbf{GSIL}$ eliminates the need for complex adversarial training in standard imitation learning, achieving lightweight and efficient fine-tuning for large language m"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2410.10093","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-14T02:21:29Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"538674512838f34c7a7a67111e8b90ca9b6ce088a396e12bc8e9903e123b5c19","abstract_canon_sha256":"0c522278e604363095572985a2ef46f17de0f88b775ec24f0f5155b770fc268d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:20:11.070303Z","signature_b64":"4vDaBVztcSscIAV3qynyCPVZw1FT1leNuKSdA7P0AqyH3M9j+hwv2gCN1exJu+1qxjV5LhOWPcJzs/e8nDCcAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9b6e77a4d6b060c2b12a2a0e7bab3a156eaf29eadbefd99146e0769962270013","last_reissued_at":"2026-07-05T09:20:11.069881Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:20:11.069881Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How to Leverage Demonstration Data in Alignment for Large Language Model? A Self-Imitation Learning Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Chao Cui, Huaisheng Zhu, Mingxiao Li, Teng Xiao, Vasant G Honavar, Yige Yuan","submitted_at":"2024-10-14T02:21:29Z","abstract_excerpt":"This paper introduces a novel generalized self-imitation learning ($\\textbf{GSIL}$) framework, which effectively and efficiently aligns large language models with offline demonstration data. We develop $\\textbf{GSIL}$ by deriving a surrogate objective of imitation learning with density ratio estimates, facilitating the use of self-generated data and optimizing the imitation learning objective with simple classification losses. $\\textbf{GSIL}$ eliminates the need for complex adversarial training in standard imitation learning, achieving lightweight and efficient fine-tuning for large language m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.10093","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2410.10093/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2410.10093","created_at":"2026-07-05T09:20:11.069938+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.10093v1","created_at":"2026-07-05T09:20:11.069938+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.10093","created_at":"2026-07-05T09:20:11.069938+00:00"},{"alias_kind":"pith_short_12","alias_value":"TNXHPJGWWBQM","created_at":"2026-07-05T09:20:11.069938+00:00"},{"alias_kind":"pith_short_16","alias_value":"TNXHPJGWWBQMFMJK","created_at":"2026-07-05T09:20:11.069938+00:00"},{"alias_kind":"pith_short_8","alias_value":"TNXHPJGW","created_at":"2026-07-05T09:20:11.069938+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.13158","citing_title":"Inverse Reinforcement Learning Meets Large Language Model Post-Training: Basics, Advances, and Opportunities","ref_index":100,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV","json":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV.json","graph_json":"https://pith.science/api/pith-number/TNXHPJGWWBQMFMJKFIHHXKZ2CV/graph.json","events_json":"https://pith.science/api/pith-number/TNXHPJGWWBQMFMJKFIHHXKZ2CV/events.json","paper":"https://pith.science/paper/TNXHPJGW"},"agent_actions":{"view_html":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV","download_json":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV.json","view_paper":"https://pith.science/paper/TNXHPJGW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.10093&json=true","fetch_graph":"https://pith.science/api/pith-number/TNXHPJGWWBQMFMJKFIHHXKZ2CV/graph.json","fetch_events":"https://pith.science/api/pith-number/TNXHPJGWWBQMFMJKFIHHXKZ2CV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV/action/storage_attestation","attest_author":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV/action/author_attestation","sign_citation":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV/action/citation_signature","submit_replication":"https://pith.science/pith/TNXHPJGWWBQMFMJKFIHHXKZ2CV/action/replication_record"}},"created_at":"2026-07-05T09:20:11.069938+00:00","updated_at":"2026-07-05T09:20:11.069938+00:00"}